Proliferating cell nuclear antigen recruits cyclin-dependent kinase inhibitor Xic1 to DNA and couples its proteolysis to DNA polymerase switching.

The Xenopus cyclin-dependent kinase (CDK) inhibitor, p27(Xic1) (Xic1), binds to CDK2-cyclins and proliferating cell nuclear antigen (PCNA), inhibits DNA synthesis in Xenopus extracts, and is targeted for ubiquitin-mediated proteolysis. Previous studies suggest that Xic1 ubiquitination and degradation are coupled to the initiation of DNA replication, but the precise timing and molecular mechanism of Xic1 proteolysis has not been determined. Here we demonstrate that Xic1 proteolysis is temporally restricted to late replication initiation following the requirements for DNA polymerase alpha-primase, replication factor C, and PCNA. Our studies also indicate that Xic1 degradation is absolutely dependent upon the binding of Xic1 to PCNA in both Xenopus egg and gastrulation stage extracts. Additionally, extracts depleted of PCNA do not support Xic1 proteolysis. Importantly, while the addition of recombinant wild-type PCNA alone restores Xic1 degradation, the addition of a PCNA mutant defective for trimer formation does not restore Xic1 proteolysis in PCNA-depleted extracts, suggesting Xic1 proteolysis requires both PCNA binding to Xic1 and the ability of PCNA to be loaded onto primed DNA by replication factor C. Taken together, our studies suggest that Xic1 is targeted for ubiquitination and degradation during DNA polymerase switching through its interaction with PCNA at a site of initiation.

The Xenopus cyclin-dependent kinase (CDK) inhibitor, p27 Xic1 (Xic1), binds to CDK2-cyclins and proliferating cell nuclear antigen (PCNA), inhibits DNA synthesis in Xenopus extracts, and is targeted for ubiquitin-mediated proteolysis. Previous studies suggest that Xic1 ubiquitination and degradation are coupled to the initiation of DNA replication, but the precise timing and molecular mechanism of Xic1 proteolysis has not been determined. Here we demonstrate that Xic1 proteolysis is temporally restricted to late replication initiation following the requirements for DNA polymerase ␣-primase, replication factor C, and PCNA. Our studies also indicate that Xic1 degradation is absolutely dependent upon the binding of Xic1 to PCNA in both Xenopus egg and gastrulation stage extracts. Additionally, extracts depleted of PCNA do not support Xic1 proteolysis. Importantly, while the addition of recombinant wild-type PCNA alone restores Xic1 degradation, the addition of a PCNA mutant defective for trimer formation does not restore Xic1 proteolysis in PCNA-depleted extracts, suggesting Xic1 proteolysis requires both PCNA binding to Xic1 and the ability of PCNA to be loaded onto primed DNA by replication factor C. Taken together, our studies suggest that Xic1 is targeted for ubiquitination and degradation during DNA polymerase switching through its interaction with PCNA at a site of initiation.
The progression of the vertebrate cell cycle is positively regulated by cyclin-dependent kinases (CDKs) associated with their cyclin partners and is negatively regulated by CDK inhibitors (reviewed in Refs. [33][34][35][36]. Cyclin E in association with CDK2 is required for the G 1 to S phase transition in mammals and for the initiation of DNA replication in Xenopus laevis, although its critical targets have yet to be identified (16,37,38). Cyclin A in mammals associates with CDK2 during G 1 and S phases, localizes to sites of DNA replication, and is required for the onset of S phase, although its specific role in S phase has not been well defined, and its critical substrates are not clear (24,25,39,40). Recent studies suggest that cyclin A activation must follow cyclin E activation and that cyclin A performs two roles during G 1 /S, the activation of DNA replication and the prevention of replication complex re-initiation (41)(42)(43)(44)(45).
The Cip/Kip family of mammalian CDK inhibitors is comprised of p21 Cip1 , p27 Kip1 , and p57 Kip2 and exhibits a broad range of inhibitory activity toward many CDK-cyclin complexes (46 -51). All Cip/Kip-type CDK inhibitors contain CDK-binding domains within their amino termini, while p21 Cip1 and p57 Kip2 also contain PCNA-binding domains within their carboxyl-terminal residues (52)(53)(54)(55)(56). In normal diploid cells, PCNA is found in multiple independent quaternary complexes with p21 Cip1 , CDKs, and cyclins A, B, and D (57,58). Before the onset of S phase, both p27 Kip1 and p21 Cip1 are targeted for ubiquitin-dependent proteolysis, while p21 Cip1 is also directly targeted for proteolysis by the proteasome (59 -68). Expression of a non-degradable p27 Kip1 mutant causes mammalian cells to arrest in G 1 , indicating that the proteolysis of p27 Kip1 is a prerequisite for the entry into S phase (62). In mammals, CDK inhibitors are targeted for proteolysis at a point before the onset of DNA replication, but it is unclear when this turnover occurs in relation to the specific molecular events of DNA replication initiation. Studying the precise timing and mechanism of CDK inhibitor proteolysis can aid in understanding the role of CDKs and CDK inhibitors in regulating the start of processive DNA replication.
In Xenopus, two CDK inhibitors have been identified that share sequence and functional similarity with both p27 Kip1 and p21 Cip1 . Xenopus inhibitor of CDK (p27 Xic1 or Xic1) and kinase inhibitor from Xenopus (p28 Kix1 or Kix1) share ϳ90% amino acid sequence identity with each other, preferentially inhibit the activity of CDK2-cyclins E and/or A, and bind all CDK-cyclins and PCNA (69,70). Xic1 and Kix1 have been shown to inhibit nuclear DNA synthesis in egg extracts through their binding to CDK2-cyclins (69,70), while Xic1 has also been shown to inhibit the replication of single-stranded DNA through its binding to PCNA (69). Studies have demonstrated that in Xenopus interphase egg extracts, Xic1 is targeted for nuclei-dependent ubiquitination and degradation in a manner that is independent of CDK2-cyclin binding and phosphorylation at CDK consensus sites (71)(72)(73)(74). Recently, two additional CDK inhibitors have been identified in Xenopus, p16 Xic2 and p17 Xic3 , and these CDK inhibitors may be orthologs of the mammalian inhibitors, p21 Cip1 and p27 Kip1 , respectively (75).
The cell-free interphase egg extract from Xenopus can recapitulate the synchronous temporal events of DNA replication initiation and elongation and can support the proteolysis of Xenopus CDK inhibitors (72,76). Using this model extract system, past studies suggest that Xic1 ubiquitination and degradation are coupled to the initiation of DNA replication, but the molecular mechanism of Xic1 proteolysis has not been determined. Our studies indicate that Xic1 is degraded during late replication initiation following DNA unwinding, RFC, and PCNA function. Importantly, we demonstrate that the ubiquitination and degradation of Xic1 is critically dependent upon the binding of Xic1 to PCNA and the ability of PCNA to be loaded onto primed DNA by RFC. Taken together, our findings suggest that Xic1 is targeted for ubiquitination and degradation when recruited to a site of initiation by PCNA, thereby coupling the proteolysis of Xic1 to DNA polymerase switching.

EXPERIMENTAL PROCEDURES
Preparation of Xenopus Interphase Egg Extracts, Embryonic Extracts, Xenopus Demembranated Sperm Chromatin, M13 Single-stranded DNA, and Methyl Ubiquitin-Xenopus egg interphase extract (low speed supernatant or LSS) and demembranated sperm chromatin were prepared as described previously (72,74,77,78). The membrane-free high speed supernatant (HSS) was prepared from fresh LSS by centrifugation at 55,000 rpm for 1 h at 4°C using a RP55S rotor and a Sorvall RC M120 centrifuge. The clear yellowish supernatant was collected and then re-centrifuged for 25 min as above. The cleared HSS was supplemented with glycerol to 4% and quick-frozen in liquid nitrogen. Degradation and replication assays were performed in the HSS either neat or diluted with XB (100 mM KCl, 0.1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM Hepes, pH 7.7) to a final protein concentration of ϳ15 mg/ml. Embryonic extracts were generated from stage 11-12 embryos as described for the LSS with the exception that the whitish layer containing nuclei, and the cytosolic clear layer were collected and pooled. Embryonic extracts were supplemented with cycloheximide immediately before use and used at a nuclei concentration of ϳ19,000 nuclei/l for degradation assays. M13mp18 single-stranded DNA (New England Biolabs) was amplified in XL1-Blue (Stratagene) as described previously (79,80) and used at a final concentration of 15 ng/l in undiluted HSS or 10 ng/l in diluted HSS. The purified single-stranded M13 DNA was a mixture of both circular and linear species. Ubiquitin (Sigma) was methylated as described previously (81) and used at a final concentration of 2 mg/ml. DNA Constructs, Site-directed Mutagenesis, and DNA Sequencing-Xic1 deletion mutants were generated by PCR mutagenesis using pCS2ϩ/Xic1-WT (72) as the template plasmid followed by subcloning into pCS2ϩ/GST (74). For the generation of GST-Xic1-(161-190), PCR primers 5Ј-TCCCCCGGGAACATCTACACAGCGCAGAAAGAG and 5Ј-CGGAATTCTCAATCAGGCTTGGCACTCAGTATCT were used, and the PCR product was cloned into the SmaI and EcoRI sites of pCS2ϩ/GST and pGEX2T (Amersham Biosciences). Xic1-WT was subcloned from the BamHI and EcoRI sites of pCS2ϩ into pGEX2T. Xic1-(161-210) was subcloned from pCS2ϩ/GST-Xic1-(161-210) as a blunt (filled-in XbaI)-BamHI fragment into the BamHI and SmaI sites of pGEX2T. Xenopus PCNA (XPCNA) was PCR-cloned from a stage 12 Xenopus embryonic cDNA library (Clontech) into the BamHI and StuI sites of pCS2ϩ using the following primers: 5Ј-CGCGGATCCATGTT-TGAGGCTCGCTTGGTGCA and 5Ј-GAAGGCCTTAAGAAGCT-TCTTCATCTTCAATCT. Point mutations in the Xic1 and XPCNA genes were generated using pCS2ϩ/Xic1 or pCS2ϩ/XPCNA plasmids, the QuikChange site-directed mutagenesis kit (Stratagene), and the following primers: Xic1-I174A, 5Ј-GAGATCACCACTCCCGCCACCG-ATTATTTCCCT and 5Ј-AGGGAAATAATCGGTGGCGGGAGTG-GTGATCTC; XPCNA-Y114A, 5Ј-CAAGAGAAAGTTTCAGACGC-TGAAATGAAGCTGATGG and 5Ј-CCATCAGCTTCATTTCAGC-GTCTGAAACTTTCTCTTG. All mutations were confirmed by DNA sequencing using the Amplicycle sequencing kit (PerkinElmer Life Sciences). For bacterial expression, XPCNA was subcloned as a BamHI and StuI fragment from pCS2ϩ/XPCNA into the BamHI and SmaI sites of pQE30 (Qiagen). XPCNA was also subcloned from pCS2ϩ/XPCNA as a BamHI and XhoI fragment into pET28a (Novogen). The construct, pCS2ϩ/Xic1-6A, was kindly provided by P. K. Jackson (73).
In Vitro Transcription and Translation-Wild-type and mutant Xic1 were in vitro transcribed and translated from the SP6 promoter in pCS2ϩ using [ 35 S]methionine (PerkinElmer Life Sciences) and the TNT-coupled reticulocyte lysate system (Promega).
Recombinant Protein Expression and Purification-GST-Xic1 fusion proteins were expressed in HB101 (for GST pull-down assays) or in BL21(pLysS)DE3 (for depletion studies). XPCNA cloned into pQE30 was expressed in M15 for antibody production, while XPCNA cloned into pET28a was expressed in BL21(pLysS)DE3 for rescue assays. GSTand His 6 -tagged proteins were purified using glutathione (Amersham Biosciences) or nickel-nitrilotriacetic acid (Qiagen)-Sepharose as described by the manufacturers. Recombinant proteins were dialyzed into XB and concentrated before use.
Degradation, DNA Replication, and Nuclei or Chromatin Spin Down Assays-Degradation, DNA replication, and nuclei or chromatin spin down assays were conducted as described previously with the following noted exceptions (74). [ 35 S]Methionine-labeled Xic1 was added to a final concentration of ϳ15 nM in degradation assays. For nuclei or chromatin spin down assays, the timing of Xic1 ubiquitination and PCNA chromatin association were performed in the absence of methyl ubiq-uitin. Additionally, demembranated sperm chromatin and [ 35 S]methionine-labeled GST-Xic1 were added simultaneously. Centrifugations were performed at 3000 ϫ g, for 5 min at 4°C using the BIOshield rotor 75006435 and Sorvall Legend RT centrifuge. The percentage of DNA replication was normalized to 100% of the control sample. Quantitation of chromatin-bound Xic1 or PCNA was determined as follows. By quantitative immunoblot analysis, the concentration of in vitro translated Xic1 was determined to be 296 nM in programmed reticulocyte lysate and a final concentration of 15 nm in the supplemented egg extract which contains 2 nM of endogendous Xic1 (determined by quantitative immunoblot analysis; Ref. 70). The concentration of endogenous PCNA in egg extract was determined to be 1 M by quantitative immunoblot analysis. Amounts of total and chromatin-bound polyubiquitinated and non-ubiquitinated Xic1 were determined by phosphorimager and quantitative immunoblot analysis. Amounts of total and chromatin-bound PCNA were determined by quantitative immunoblot analysis. Experiments were generally conducted at least three or more times with the results indicating the same trend but with absolute values potentially varying between 5 and 10% due to extract to extract or experimental variations.
Depletion and Rescuing-GST-Xic1-(161-190) was coupled to glutathione-Sepharose 4B to a final concentration of 80 M and then incubated in the egg extract for 1 h at 4°C. The depleted extracts were then separated from the beads by centrifugation at 4°C. Control depletions were performed using GST coupled to glutathione-Sepharose beads. The depleted extracts were rescued with XB buffer or bacterially expressed and purified wild-type or Y114A mutant His 6 -T7-XPCNA.

RESULTS
The Temporal Requirements for Xic1 Proteolysis: Xic1 Is Degraded during Late Initiation-Previous studies have characterized Xic1 proteolysis in membrane-containing cytosolic egg extracts often called the low speed supernatant, or LSS, supplemented with sperm chromatin or in membrane-free nucleoplasmic extracts called NPE (71)(72)(73)(74)83). The LSS supports nuclei formation and all the temporal events of DNA replication including pre-RC and pre-IC formation, DNA unwinding, DNA polymerase switching, and elongation (76,77). The NPE supplemented with chromatin supports the events of DNA replication in the absence of nuclei formation (84). High speed centrifugation of the LSS results in a soluble, cytoplasmic, membrane-free extract called the high speed supernatant or HSS (77,85). Due to the absence of nuclear membrane precursors, the HSS will not support the formation of nuclei and thus will not support the events of pre-RC and pre-IC formation or origin unwinding (85). However, the HSS will support DNA replication processes that are not dependent upon origin unwinding, namely single-stranded DNA synthesis, which requires only the late initiation events of primer synthesis, DNA polymerase switching, and elongation (85). Thus, the HSS has been used as a model for the study of singlestranded DNA replication and the activities of DNA polymerase ␣-primase, RFC, PCNA, and DNA polymerase ␦.
To determine whether Xic1 degradation was supported by the singlestranded DNA replication events supported in the HSS, we used the HSS supplemented with single-stranded DNA. We found that while Xic1 was efficiently degraded in the LSS in the presence of either single-or double-stranded DNA, Xic1 was only degraded in the HSS in the presence of single-stranded DNA (Fig. 1, A and B). This result suggests that Xic1 proteolysis is triggered by the presence of single-stranded DNA or unwound DNA and is triggered during DNA polymerase switching or elongation, replication events supported by the HSS.
We next addressed whether the HSS extract system recapitulates the same regulation of Xic1 proteolysis previously demonstrated in the LSS and the NPE. Because we use both the LSS and the HSS in the following studies to examine Xic1 proteolysis, we tested whether the mechanism of Xic1 degradation is the same in the LSS, the NPE, and the HSS. In the LSS containing nuclei and the NPE containing chromatin, Xic1 proteolysis is not dependent upon CDK2-cyclin binding (71,74). Additionally in the LSS, Xic1 proteolysis is not dependent upon phosphorylation at six consensus CDK2 sites but is critically dependent upon basic residues within amino acids 180 -183 (73,74). Using the HSS supplemented with single-stranded DNA, we examined the degradation of three key Xic1 mutants: ckϪ (unable to efficiently bind CDK2-cyclins) (74), 6A (substituted at all six consensus CDK2 phosphorylation sites) (73), and NLS2 (mutated at basic amino acids 180 -183) (74). Identical to published results obtained in the LSS and the NPE (71,73,74), our results demonstrate that Xic1 proteolysis in the HSS with single-stranded DNA is not dependent upon CDK2-cyclin binding or phosphorylation by CDK2-cyclin but is dependent upon the small domain within amino acids 180 -183 ( Fig. 1C and Fig. 4C, right panel). To further study whether Xic1 proteolysis in the LSS and the HSS is mechanistically the same, we identified the minimal domain that was necessary and sufficient for Xic1 proteolysis in both extracts. Our studies demonstrate that in both the LSS and the HSS, the carboxyl-terminal 50 amino acids of Xic1 are necessary and sufficient for efficient Xic1 proteolysis ( Fig. 1D; Ref. 86). Moreover, we find that 30 amino acids (161-190) of Xic1 are sufficient for Xic1 ubiquitination in the HSS and the LSS (Fig. 1D; Ref. 86), similar to a finding by You et al. (71) using the NPE. These results suggest that Xic1 degradation occurs during the late initiation events of primer synthesis, DNA polymerase switching, or elongation and that the mechanism of Xic1 proteolysis in the LSS, the HSS, and the NPE is the same.
Xic1 Degradation Requires the Synthesis of RNA and DNA Primers-The HSS supports Xic1 proteolysis in the presence of single-stranded DNA, but not double-stranded DNA, implying that beyond the require-ment for DNA, Xic1 degradation may require DNA unwinding, which results in the generation of single-stranded DNA. This result also implies that Xic1 proteolysis occurs during either late initiation or elongation, when origin unwinding would be completed. To determine the timing of Xic1 proteolysis more precisely, we added inhibitors of different steps of late replication initiation and then tested whether Xic1 proteolysis was supported in the treated extracts. CDK2-cyclin E activity is required prior to origin unwinding during late replication initiation (16,17,87). Our studies show that the addition of the CDK inhibitor, roscovitine, blocks Xic1 degradation in the LSS, a result previously reported by Furstenthal et al. (73), but does not inhibit Xic1 degradation in the HSS ( Fig. 2A). This result suggests that Xic1 degradation is dependent upon CDK2-cyclin-mediated DNA unwinding but does not require CDK2-cyclin kinase activity once the DNA has been unwound. This indicates that Xic1 degradation occurs after the CDK2-cyclin E requirement and DNA unwinding.
Following origin unwinding, a RNA primer is synthesized by the primase subunit of the DNA polymerase ␣-primase complex (23). To inhibit RNA primer synthesis we added the RNA polymerase inhibitor, actinomycin D, and then measured Xic1 proteolysis. Our results indicate that at concentrations of actinomycin D that inhibit ϳ95% of DNA replication in the LSS and the HSS, we observed a marked reduction in Xic1 degradation compared with controls ( Fig. 2A). The inhibition of Xic1 degradation by actinomycin D was consistently less in the HSS, perhaps due to increased priming of single-stranded DNA. This result indicates that Xic1 proteolysis requires RNA priming by Primase and suggests that Xic1 is degraded after the synthesis of RNA primers.
After RNA priming, DNA polymerase ␣ synthesizes a DNA primer that is recognized by RFC to load PCNA onto primed single-stranded DNA (23). Aphidicolin inhibits DNA primer synthesis by DNA polymerase ␣ and prevents the recruitment of RFC and PCNA to DNA. When aphidicolin was added to the LSS or the HSS at concentrations that inhibited ϳ96% of DNA replication, Xic1 proteolysis was significantly inhibited (Fig. 2B). Furthermore, the addition of a blocking antibody directed against DNA polymerase ␣ (SJK132) (88,89) also significantly inhibited the degradation of Xic1 in the LSS and the HSS at concentrations that inhibited at least 80% of DNA replication (Fig. 2, C and D). These studies suggest that Xic1 proteolysis requires the activity of DNA polymerase ␣ and indicates that Xic1 proteolysis occurs after the synthesis of DNA primers. These studies also suggest that Xic1 degradation requires the presence of primed single-stranded DNA.
Xic1 Degradation Requires the Function of RFC-Following origin unwinding and primer synthesis, RFC recognizes primed singlestranded DNA and loads PCNA at a site of replication initiation (23). To further study the timing of Xic1 proteolysis, we blocked RFC function using three different antibodies against the large subunit of human RFC and then tested whether Xic1 was degraded in the treated extracts ( Fig.  3A). At concentrations that blocked ϳ98% of DNA replication, two monoclonal antibodies directed against the large subunit of RFC significantly inhibited Xic1 proteolysis in the LSS compared with the control antibody ( Fig. 3B; Ref. 82). Furthermore, a rabbit antibody directed against the large subunit of RFC also strongly inhibited Xic1 proteolysis in the LSS and moderately inhibited DNA replication and Xic1 proteolysis in the HSS (Fig. 3C). The limited inhibition of DNA replication observed in the HSS with the RFC antibody is most likely due to the RFC-independent DNA replication of linear single-stranded DNA. Linear single-stranded DNA can be replicated by a PCNA-dependent, but RFC-independent, pathway (90). This result indicates that in an RFCdependent replication system, Xic1 proteolysis requires the function of RFC. These studies also suggest that Xic1 is targeted for degradation during late initiation following RFC function.
Xic1 Proteolysis Is Dependent upon PCNA Binding-Our studies suggest that Xic1 may be degraded during DNA polymerase switching and requires the function of RFC and perhaps PCNA. Our previous studies indicated that residues 161-210 of Xic1 were necessary and sufficient for the ubiquitination and degradation of Xic1 and that amino acids 180 -183 were critical determinants of Xic1 proteolysis (74,86). Analysis of this carboxyl-terminal region of Xic1 reveals conserved residues shown to be important for p21 Cip1 binding to PCNA (Fig. 4A) (69, 91). Taken together, we wondered whether PCNA might play a critical role in modulating Xic1 ubiquitination and degradation. We first generated a point mutation in Xic1 at amino acid 174 that we predicted would disrupt Xic1 binding to PCNA based on the crystal structure solved for p21 Cip1 in association with PCNA (91,92). We then measured the ability of this and several other Xic1 mutants defective for PCNA binding to be degraded. Our results show a striking correlation between the ability of a Xic1 mutant to bind endogenous PCNA in egg extract and its ability to be degraded in the LSS or the HSS (Fig. 4, B and C; Ref. 74). This result also strongly suggests that the Xic1 mutant, NLS2( 180 KRKK 183 to 180 ARAA 183 ), was not degraded in previous studies because it was defective for PCNA binding (74). All of the Xic1 mutants tested bound CDK2-cyclin E as efficiently as wild-type Xic1 (data not shown; Ref. 74). Significantly, although the Xic1 I174A mutant efficiently localized to the nucleus and bound to chromatin, it was not ubiquitinated (Fig. 4D). Furstenthal et al. (73) proposed that Xic1 is recruited to chromatin through its interaction with CDK2-cyclin E and that this interaction is critical for its ubiquitination and degradation. However, our studies demonstrate that Xic1 binding to CDK2-cyclin E and to chromatin are not sufficient for Xic1 ubiquitination. Instead, our studies indicate that Xic1 must bind PCNA to be ubiquitinated and degraded.
Trimeric PCNA Recruits Xic1 to Chromatin where It Is Targeted for Ubiquitination-Our studies suggest that PCNA binding to Xic1 is required for Xic1 proteolysis, but they do not exclude the possibility that an alternate protein, which binds to the same COOH-terminal PCNAbinding region of Xic1, may be the critical mediator of Xic1 degradation. To address this possibility, we depleted the endogenous PCNA from the LSS and the HSS using a GST-Xic1-(161-190) polypeptide containing the PCNA-binding domain (Fig. 5A) (93). Depletion of PCNA prevented Xic1 degradation in the LSS and the HSS, while Xic1 was efficiently degraded in control-depleted extracts (Fig. 5, B and C). Immunoblot analysis confirmed that PCNA was efficiently depleted from extracts, while the concentration of CDK2-cyclin E was not affected ( Fig. 5B; data not shown). Importantly, the addition of recombinant PCNA alone was able to fully rescue Xic1 proteolysis in depleted extracts (Fig. 5C). This result demonstrates that although other proteins may bind to Xic1-(161-190) and may be depleted along with PCNA, only the addition of PCNA is required to restore Xic1 proteolysis in depleted extracts. Moreover, because the majority of Xic1 proteolysis was blocked by the removal of PCNA, this implies that PCNA plays a critical role in mediating Xic1 proteolysis in the egg, perhaps through the recruitment of Xic1 to chromatin at a site of initiation.
To test this possibility directly, we generated a mutant of Xenopus PCNA that could not be loaded onto DNA (90) and analyzed the ability of this mutant to mediate Xic1 proteolysis. We based our mutagenesis strategy on studies performed on the highly conserved human PCNA protein, since Xenopus and human PCNA share ϳ90% amino acid identity. Crystal structure studies demonstrate that human PCNA forms a homotrimeric ring that functions as a sliding clamp encircling a duplex of DNA (91). Mutation of the conserved tyrosine 114 to alanine impairs trimer formation of human PCNA and prevents its RFC-mediated loading onto DNA (90). We generated the Y114A mutation in Xenopus PCNA and tested its ability to rescue Xic1 proteolysis in PCNA-depleted extracts. While the wild-type PCNA fully restored Xic1 proteolysis, the Y114A PCNA mutant did not (Fig. 5C). Binding studies demonstrated that both the wild-type and Y114A PCNA proteins could bind Xic1 efficiently, although the latter was somewhat reduced (Fig. 5D).
These studies demonstrate that the binding of Xic1 to PCNA is necessary, but not sufficient, for Xic1 proteolysis. These studies also suggest that beyond the binding of Xic1 to PCNA, Xic1 degradation may be dependent upon the ability of PCNA to be loaded onto DNA by RFC. To study this hypothesis, we examined the chromatin binding of PCNA, Xic1, and RPA in egg extracts during DNA replication initiation. Our studies show that during replication initiation, endogenous RPA associates with chromatin 30 -60 min following the addition of sperm chro- matin to the LSS (Fig. 6, A and B). Endogenous PCNA associates with chromatin following RPA binding to DNA and the association of PCNA on chromatin parallels the appearance of ubiquitinated Xic1 on chromatin (Fig. 6B). At 60 min when PCNA binding to chromatin reaches its peak, ϳ80% of Xic1 has been ubiquitinated and degraded (Fig. 6B). We quantitated the amounts of PCNA (67 fmol) and mono-, di-, or polyubiquitinated [ 35 S]GST-Xic1 (9.7 fmol) that were bound to chromatin at the 30-min time point and found that ϳ7 times more moles of PCNA were bound to chromatin compared with Xic1 (Fig. 6B, left panels, lane  2). In PCNA-depleted extracts, no chromatin-bound PCNA is detected, and Xic1 ubiquitination is severely reduced, while in depleted extracts supplemented with recombinant wild-type PCNA, chromatin-bound PCNA is observed and is accompanied by chromatin-bound ubiquitinated Xic1 (Fig. 6, A and C). PCNA-depleted extracts supplemented with the Y114A mutant of PCNA exhibit virtually no chromatin-bound PCNA or ubiquitinated Xic1 (Fig. 6, A and C). These results show a perfect correlation between the binding of PCNA to chromatin and Xic1 ubiquitination. Taken together with our previous findings, we believe that these studies suggest that Xic1 proteolysis is triggered by the loading of Xic1-bound PCNA to a site of initiation during DNA polymerase switching.
Xic1 Proteolysis in the Egg and in Gastrulating Embryos Is Mediated Predominantly through PCNA Binding-Our results indicate that Xic1 proteolysis in the Xenopus egg occurs predominantly by a pathway dependent upon PCNA, but it is unclear whether this pathway degrades Xic1 during a somatic cell cycle when Xic1 is much more highly expressed. To study this, we generated peptides containing the PCNAbinding domain of Xic1 fused to GST and added the peptides to extracts to disrupt the interaction between Xic1 and PCNA. In the peptidetreated LSS and HSS extracts, Xic1 proteolysis was significantly inhibited (Fig. 7A). This result is consistent with a report which demonstrated that the addition of a Xic1-(162-192) fusion protein with green fluorescent protein inhibits Xic1 degradation in the NPE (71). Additionally, our studies and the studies of You et al. (71) demonstrate that the addition of Xic1 COOH-terminal peptides at ϳ3 M does not significantly inhibit DNA replication (data not shown), even though it inhibits Xic1 proteolysis. The inhibition of Xic1 degradation by Xic1 COOH-terminal peptides was fully reversed in the LSS and partially reversed in the HSS upon the addition of recombinant wild-type Xenopus PCNA (data not shown).
In the egg, very little Xic1 protein is detected (ϳ2 nM), but these levels increase dramatically after the mid-blastula transition when the asynchronous somatic cell cycle begins to be established (Fig. 7B) (data not shown; Ref. 70). We generated extracts from stage 11-12 gastrulating embryos and treated these extracts with the Xic1 COOH-terminal peptides. Our results indicate that the proteolysis of Xic1 in stage 11-12 extracts is markedly inhibited by the peptides that disrupt the binding of Xic1 to PCNA (Fig. 7C). In addition, the proteolysis of a Xic1 point mutant defective for PCNA binding (I174A) was reduced in the stage 11-12 extract (Fig. 7D). These studies suggest that the predominant Xic1 degradation pathway present during the Xenopus embryonic and somatic cell cycles is the one mediated by PCNA.  Remaining). DNA replication at the 3-h time point was normalized to 100% of the replication observed with the appropriate control buffer (% DNA Replication). D, Xic1 and PCNA binding assay. Bacterial lysate containing GST or GST-Xic1 WT was bound to glutathione-Sepharose and then incubated in LSS that was either control-depleted with GST bound to glutathione-Sepharose (CTRL DEPL) or PCNA-depleted with GST-Xic1-(161-190) bound to glutathione-Sepharose (XPCNA DEPL). PCNA-depleted LSS was supplemented before binding to beads with XB-(BUFFER) or recombinant WT or Y114A His 6 -T7-tagged Xenopus PCNA (WT or Y114A). Washed beads were eluted and analyzed by SDS-PAGE and immunoblotting using antibody to PCNA. Ten percent of the input LSS is shown (10% INPUT) and PCNA binding is indicated as a percentage (% BINDING) of the input PCNA. Arrows denote either endogenous PCNA (XPCNA) or recombinant PCNA (His6-T7-XPCNA). The PCNA-depleted LSS used in Fig. 5D is the same depleted LSS used in Fig. 5, B and C, and Fig. 6A.

DISCUSSION
Our studies suggest a model for the proteolysis of Xic1 in the egg extract and in the somatic cell (Fig. 8). We have found that although Xic1 may be recruited to chromatin through an interaction with CDK2cyclin E and Cdc6 (73,94), the Cdc6-mediated association of Xic1 to chromatin is not sufficient to trigger Xic1 ubiquitination and the interaction of Xic1 with CDK2-cyclin is dispensable for Xic1 degradation (71,74). Our findings suggest that Xic1 proteolysis occurs after the formation of the pre-RC, after DNA unwinding, and after the replication requirements for RPA, DNA polymerase ␣-primase, RFC, and PCNA. Xic1 proteolysis is absolutely dependent upon the presence of DNA and the binding of Xic1 to trimeric PCNA, suggesting that only PCNA that is competent for loading to a site of initiation by RFC can mediate the ubiquitination of Xic1. This also implies that Xic1 is only ubiquitinated when bound to chromatin and suggests that chromatin  serves to nucleate Xic1 and the Xic1 ubiquitination machinery. Based on studies of mammalian p21 Cip1 and PCNA, it is predicted that the binding of Xic1 to PCNA may block the recruitment of DNA polymerase ␦ to a site of initiation (91). It is still unclear whether the proteolysis of Xic1 is required for the recruitment of DNA polymerase ␦ or whether processive DNA polymerases may play a role in triggering the ubiquitination of Xic1.
The results on the timing of Xic1 proteolysis are somewhat controversial. Furstenthal et al. (73) report that Xic1 is degraded following pre-RC formation but before DNA polymerase ␣ activity. Their studies are based on the detection of ubiquitinated Xic1 associated with chromatin in the presence of 40 g/ml aphidicolin (73). We also find that in extracts containing membranes such as the LSS, some ubiquitinated Xic1 can be observed in the chromatin fraction at aphidicolin concentrations below 50 -100 g/l, while in different extracts, aphidicolin concentrations that block Xic1 degradation may vary by 2-3-fold (data not shown). In contrast, we do not observe this variation in the HSS. You et al. (71) show that Xic1 proteolysis occurs during late initiation following Cdc45 recruitment but before RPA or DNA polymerase ␣ function. These studies are based on the observation that Xic1 is proteolyzed in RPA-depleted extracts and in the presence of 1 mg/ml SJK132 IgG directed against DNA polymerase ␣, although the kinetics of Xic1 degradation are noticeably slowed upon both treatments (71). Similar to You et al. (71), we also observe that the kinetics of Xic1 degradation are slowed in the presence of 1 mg/ml SJK132 IgG (data not shown). However, our studies show that at higher concentrations of SJK132 between 2 and 3 mg/ml, Xic1 degradation is inhibited. Because we study Xic1 degradation in both the LSS and the HSS, we believe that our findings provide a more reliable examination of the timing of Xic1 proteolysis.
Our finding that the ubiquitination and degradation of Xic1 occurs during DNA polymerase switching is unexpected, since it has been predicted that Xic1 targets CDK2-cyclin E prior to DNA unwinding. Furstenthal et al. (73) have reported that Xic1 bound to CDK2-cyclin E can be recruited to chromatin through the association of cyclin E with Cdc6, although high levels of Xic1 block the association of CDK2-cyclin E to chromatin via Cdc6. The RXL motifs of Cdc6 and Xic1 both bind to the MRAIL domain of cyclin E, indicating that the binding of Cdc6 and Xic1 to cyclin E is mutually exclusive (94). Therefore, Xic1 recruited to chromatin by a CDK2-cyclin E-Cdc6 interaction would be associated to chromatin only through its interaction with CDK2, and it is unknown whether this interaction is sufficient to fully inhibit kinase activity. Consequently, the degradation of Xic1 associated with CDK2-cyclin E and Cdc6 on chromatin prior to DNA unwinding may be dispensable. Alternatively, during the somatic cell cycle when Xic1 is more highly expressed, Xic1 proteolysis may be required prior to DNA unwinding in a PCNA-independent manner as well as during DNA polymerase switching by a PCNA-dependent manner.
Based on our findings of the timing and mechanism of Xic1 ubiquitination and degradation, we propose that during the somatic cell cycle, the PCNA-dependent degradation of Xic1 during DNA polymerase switching may activate CDK2-cyclin A and/or PCNA. Consistent with this model, we have observed in the LSS that the addition of recombinant wild-type Xic1 to 60 nM, a concentration that more closely mimics the Xic1 concentration in a somatic cell, delayed the entry into S phase, while the addition of the non-degradable I174A Xic1 mutant blocked S phase entry, suggesting that the PCNA-dependent proteolysis of Xic1 is important for the onset of elongation. 4 The degradation of Xic1 by this mechanism would couple the activation of a CDK2-cyclin with the start of processive DNA replication and would ensure that any process that halts S phase before DNA polymerase switching would result in the stabilization of Xic1. In Xenopus, cyclin A2 most closely resembles human cyclin A in sequence and expression, although the function of cyclin A2 is not well characterized (95).
Although we believe our studies in the embryonic and gastrulation stage extracts suggest that Xic1 binding to PCNA is important for Xic1 turnover, it is still unclear exactly how Xic1 proteolysis is regulated during the somatic cell cycle. Our studies and the studies of other groups have primarily used egg extracts and exogenously added Xic1 to characterize the regulation of Xic1 turnover (71)(72)(73)(74)83). Notably, we have observed that endogenous Xic1 is efficiently degraded in the LSS in the presence of nuclei, demonstrating that the proteolysis of Xic1 is not just an artifact of exogenously added Xic1 (86). The requirement for added nuclei at a ratio of ϳ3200 -5000 sperm to 1 l of cytoplasm may indicate that the degradation of endogenous Xic1 is not normally observed in the embryo until around the start of the mid-blastula transition (96). Studies have demonstrated that Xic1 plays an important role in the differentiation of neural and muscle cells, and it is possible that during differentiation, Xic1 protein levels may be modulated by protein stability (97)(98)(99)(100). Future studies that examine endogenous Xic1 turnover in the somatic cell and during cellular differentiation will be required to fully understand the function and regulation of Xic1 and its targets.
It is possible that some of our experimental conditions in the LSS activated a DNA replication or DNA damage checkpoint that inhibited 4 L.-C. Chuang and P. R. Yew, unpublished observation. . We propose that the ubiquitination (Ub) of Xic1 occurs at a site of initiation and is mediated by the loading of trimeric PCNA and associated Xic1 to chromatin. In a somatic cell, we postulate that the proteolysis of Xic1 couples the activation of CDK2-cyclin A and/or PCNA with DNA polymerase switching. the proteolysis of Xic1 through an ATR (ATM and Rad3-related kinase) or ATM (ataxia telangiectasia mutated) and Chk1/Chk2 pathway (101)(102)(103)(104)(105)(106). One possibility is that ATR phosphorylation of Xic1 mediates its stability during a checkpoint. Xic1 contains one consensus ATR site at threonine 163, but our findings demonstrate that mutation of this residue has no effect on Xic1 proteolysis in the absence or presence of a checkpoint. 5 Alternatively, Chk1/Chk2 phosphorylation by ATR/ATM could result in the inhibition of CDK2-cyclin activation (105)(106)(107)(108)(109), thereby potentially inhibiting Xic1 proteolysis in the LSS. However, all of our studies were also performed in the HSS where CDK2 activity is dispensable for DNA replication and Xic1 degradation (87). Thus, our studies using the HSS are not sensitive to the effects of ATR/ATM checkpoint pathways that target CDK2-cyclin activity. Furthermore, the studies of the Xic1 I174A mutant defective for PCNA binding and the inhibition of Xic1 proteolysis through the addition of Xic1 peptides that disrupt the Xic1-PCNA interaction were not performed under checkpoint conditions, since DNA replication was not inhibited. Therefore, we believe our findings support the model that Xic1 proteolysis occurs during DNA polymerase switching through the interaction of Xic1 with PCNA on chromatin.
Because the E3 SCF components, Cul1 and p19 Skp1 , have been shown to be associated with chromatin (73), it is predicted that Xic1 is targeted for ubiquitination by a chromatin-bound SCF complex when associated at a site of initiation. However, to date, the machinery responsible for Xic1 ubiquitination, including its putative F-box protein, have yet to be identified. Additionally, although immunodepletion of Cdc34 blocks Xic1 degradation in the LSS with sperm chromatin (72), surprisingly, immunodepletion of Cdc34 from the HSS does not inhibit Xic1 degradation in the presence of single-stranded DNA. 6 This indicates that a Cdc34 requirement for DNA replication initiation exists and occurs before the timing of Xic1 degradation, implying that Cdc34 either is not the E2 for Xic1 or is not the only E2 for Xic1. It is also unclear whether the recruitment of Xic1 to a site of initiation by PCNA is sufficient to trigger Xic1 ubiquitination or whether additional signals may be required. If Xic1 is truly targeted for ubiquitination by a SCF complex, then it is predicted that Xic1 phosphorylation may play a role in the binding of Xic1 to its cognate F-box protein. However, our previous studies suggest that phosphorylation of Xic1 is not critical for its ubiquitination or degradation in the egg extract (86). During a somatic cell cycle when Xic1 is more highly expressed, the phosphorylation of Xic1 may play an important role in regulating the stability of Xic1. Studies have demonstrated that CDK2-cyclin A associates with SCF proteins, p45 Skp2 and p19 Skp1 , in a manner that is mutually exclusive of its binding to p21 Cip1 (110,111). Interestingly, CDK2 has been shown to bind directly to PCNA suggesting the possibility that PCNA may directly recruit to chromatin both a CDK inhibitor and a CDK2-cyclin bound to p45 Skp2 and p19 Skp1 (112,113).
Our studies suggest that the PCNA-dependent degradation pathway, which mediates Xic1 proteolysis in Xenopus, might also regulate the stability of the mammalian CDK inhibitors, p21 Cip1 and p57 Kip2 , that associate with PCNA. Currently, studies have demonstrated that p21 Cip1 protein turnover is mediated by the proteasome via ubiquitindependent and ubiquitin-independent pathways, but no studies have demonstrated a dependence on PCNA binding for p21 Cip1 turnover in somatic cells (66, 67, 114 -116). Neither p27 Kip1 nor p21 Cip1 are targeted for proteolysis in the egg extract, suggesting some evolutionary divergence between the regulation of mammalian and non-mammalian ver-tebrate CDK inhibitors. 7 Further studies examining the proteolysis of PCNA-binding CDK inhibitors in the context of DNA replication initiation can help to determine whether the PCNA-dependent pathway, which regulates Xic1 turnover, is a universal mode of regulation in other vertebrates.